How to build high sensitivity+high voltage function gen

[EE4life]

Hi all,

I am in need to create a system with which can output a AC signal of high voltage and also have good resolution at low voltages.

Currently, I am using a function generator with an 100X amplifier connected to a potentiometer. When I want to use low voltage I turn the potentiometer to a high resistance, and vice versa.

I want to use this equipment with automated control using labview to drive a piezo at high (off resonance) and low voltage (resonance). I would like for the resistance be changed simply using electronic switches.

I was thinking about using a MOSFET, but I do not think the voltage may be too large (100 volts) for a MOSFET.

Do you guys have any advice on my issue?

Thanks in advance.

 

[berkeman]

Hi all,

I am in need to create a system with which can output a AC signal of high voltage and also have good resolution at low voltages.

Currently, I am using a function generator with an 100X amplifier connected to a potentiometer. When I want to use low voltage I turn the potentiometer to a high resistance, and vice versa.

I want to use this equipment with automated control using labview to drive a piezo at high (off resonance) and low voltage (resonance). I would like for the resistance be changed simply using electronic switches.

I was thinking about using a MOSFET, but I do not think the voltage may be too large (100 volts) for a MOSFET.

Do you guys have any advice on my issue?

Thanks in advance.

Welcome to the PF.

What frequency range do you need? I'm assuming that the load impedance is fairly high, right? Do you need 50 Ohm output impedance, or can you live with a higher output impedance? What is your definition of "good resolution"? If you based this generator on a 12 bit full scale DAC, would that give you the resolution you need?

 

[EE4life]

Thank for your reply.

The frequency range is 10 KHz-200 KHz. At resonance the load impedance is around 50 Ohms and at antiresonance the load impedance is around 50 KOhms. A higher output impedance is okay. The resolution should be within 1/100 of the voltage value at least, so I guess that is achievable in any case.

(approximate calculations)
To drive at antiresonance (high voltage), we need 100V. So with setting the function generator to 10 volts, X100,=1000V. The potentiometer needs to be at 5 KOhms.
For resonance, assuming we need around 1 volts drive, with the same setup, if we reduce the function generator to 1v, we get 100V out of the amplifier. Since the potentiometer (5 KOhm) is in series with the piezo device the piezo is driven at 1 volt.

In this example we have placed a large externally placed impedance load to make the voltage right. I would like to get around this somehow.

 

[berkeman]

Thank for your reply.

The frequency range is 10 KHz-200 KHz. At resonance the load impedance is around 50 Ohms and at antiresonance the load impedance is around 50 KOhms. A higher output impedance is okay. The resolution should be within 1/100 of the voltage value at least, so I guess that is achievable in any case.

(approximate calculations)
To drive at antiresonance (high voltage), we need 100V. So with setting the function generator to 10 volts, X100,=1000V. The potentiometer needs to be at 5 KOhms.
For resonance, assuming we need around 1 volts drive, with the same setup, if we reduce the function generator to 1v, we get 100V out of the amplifier. Since the potentiometer (5 KOhm) is in series with the piezo device the piezo is driven at 1 volt.

In this example we have placed a large externally placed impedance load to make the voltage right. I would like to get around this somehow.

Instead of putting a resistance in series to alter the amplitude, why not just adjust the amplitude of the signal generator? Or even put a variable gain amplifier between the signal source and the load?

 

[EE4life]

Variable gain amplifiers of this sort would be expensive given the large voltage and modest power. Adjusting the signal generator won't work because a it is difficult to get a 1V-150V output with only a 100m to 10V signal generator without the large series impedance. The max power need by the piezo is around 200mW.

 

[EE4life]

Do you have a suggestion for the variable amplifier?

 

[berkeman]

Can you say more about what you are trying to achieve with this test setup? It would seem that having a varialble output impedance for the waveform source would affect the results of the piezo testing. Is that not the case? What are you looking for in your piezo measurements?

Another alternative would be to use two different signal sources for the high-voltage and low-voltage measurements. You could switch between them using a relay, for example.

 

[NascentOxygen]

Do you have a suggestion for the variable amplifier?

Does this spark any ideas?
http://www.electronics-lab.com/blog/?p=21644

and another: http://m.electronicdesign.com/analo...-piezoelectric-motor-using-low-voltage-op-amp

 

[EE4life]

I sweep the frequency (slowly) and then I will measure the impedance across the the piezo. I will measure the voltage across the piezodevice and I also am measuring the current, therefore I can measure impedance.

So, essentially I am making an hp analyzer, but I am using higher voltage. Also, using the labview program I am also doing constant current and constant electrical power frequency sweeps. With the impedance I am measuring the quality factor.

 

[berkeman]

I sweep the frequency (slowly) and then I will measure the impedance across the the piezo. I will measure the voltage across the piezodevice and I also am measuring the current, therefore I can measure impedance.

So, essentially I am making an hp analyzer, but I am using higher voltage. Also, using the labview program I am also doing constant current and constant electrical power frequency sweeps. With the impedance I am measuring the quality factor.

Are you measuring the phase of the impedance as well? An HP impedance analyzer measures complex impedance...

 

[EE4life]

Yes, I am also measuring phase between the voltage and current. This would change with the pot resistance...wouldn't it?
-----
How much power can I get out the op amp? I think that is for a low power device. The people my lab before me are/were using something like (big box type amplifer) http://www.ferrodevices.com/1/297/files/Precision_200V_HighSpeed_Ex_A3.pdf
But, perhaps they were overdoing it because the current and demands are not very high, although the voltage might be.
-----
Anyways, a relay between two function generators seems like an easy and simple solution. Do you have a suggestion for a relay for my application?

 

[vk6kro]

If you drive an amplifier which has a gain of 100 with a 1 volt signal, you do not get 100 volts out, usually.

This is because the amplifier will give distorted output at a much lower voltage. Most amplifiers would not be capable of giving 100 volts peak to peak output.

If you drive it with 50 mV peak to peak, you might get 5 volts output, but, depending on the amplifier, driving it with much more than this will usually give output which is distorted and nowhere near 100 volts in amplitude.

If you just want to test a piezo device, place it in series with a 10k resistor across the output of a signal generator, and measure the voltage across the resistor with an oscilloscope as you vary the frequency of the signal generator.

You can observe the output of the signal generator with the oscilloscope at the same time. So, you can measure phase differences, if any, between input and output of the test setup.

You can also adjust the output of the signal generator so that it remains constant at different frequencies.

For convenience, you would place the resistor at the ground end of the series circuit. This allows you to connect an oscilloscope to it easily.

 

[EE4life]

If you drive an amplifier which has a gain of 100 with a 1 volt signal, you do not get 100 volts out, usually.

This is because the amplifier will give distorted output at a much lower voltage. Most amplifiers would not be capable of giving 100 volts peak to peak output.

If you drive it with 50 mV peak to peak, you might get 5 volts output, but, depending on the amplifier, driving it with much more than this will usually give output which is distorted and nowhere near 100 volts in amplitude.

If you just want to test a piezo device, place it in series with a 10k resistor across the output of a signal generator, and measure the voltage across the resistor with an oscilloscope as you vary the frequency of the signal generator.

You can observe the output of the signal generator with the oscilloscope at the same time. So, you can measure phase differences, if any, between input and output of the test setup.

You can also adjust the output of the signal generator so that it remains constant at different frequencies.

For convenience, you would place the resistor at the ground end of the series circuit. This allows you to connect an oscilloscope to it easily.

I am not clear on why are suggesting to test the voltage over the 10k resistor. I am measuring current using a current probe.

By the way, I want to make a clarification. The potentiometer is between the function generator and the amplifier, so it does not add to the impedance of the piezoelectric device. I have attached a crude picture of my setup. Again, I use the voltage and current to calculate impedance, from which I get the quality factor (from 3db bandwidth). I also calculate phase angle between the voltage and current to get the power.

 

[vk6kro]

If you apply a steady voltage across the series Piezo--resistor combination, then the piezo is protected from excess drive and the voltage across the resistor depends on the impedance of the piezo.

Some piezo devices are rugged, but most can be damaged by excess drive.

Doing it this way, you don't need current probes or extra amplifiers.
As mentioned above, signal generators have panel controls that control output amplitude. You don't need any external potentiometers.

 

[EE4life]

Will the series resistor mess up the resonance characteristics of the piezo (typically we just measure a plate or disk of piezoelectric material, so it not really a actual device)?

At resonance the piezo impedance is between 30-100 Ohms (not kohms) and at antiresonance the impedance is around 50k. What I really want to find out is the impedance of the piezo across a frequency sweep from resonance to antiresonance.

 

[vk6kro]

You can adjust the value of the resistor to suit your device. As long as you know the resistance of the resistor, you can calculate the resistance of your piezo sample.

Possibly, 1000 ohms would be better in your case. You can easily try it and do a frequency sweep.
Ideally, you should try to have results that are varying with frequency, not staying near zero volts or near the maximum voltage.

Watch to see if the voltage across the resistor stays in phase with the input voltage. If it doesn't, you will need to use vector diagrams to calculate the impedance of the sample.

I wouldn't expect the resistor to affect the resonance frequency much or at all. You can easily find out by trying different resistors as you approach resonance.

 

[meBigGuy]

All this talk of series resistances and X100 gain amplifiers that distort is just confusing the issue.

The two circuits in NO's post (#8) will do this right. Using an input pot will give you whatever range you want.

Why wouldn't you just do it that way? Am I missing something?

 

[vk6kro]

I don't know. Are you?

To measure the resonant frequency of a "piezo" you don't need to generate high voltages as the early posts suggested.
The voltage from a signal generator is adequate.
That is where the series resistor comes in.
You compare two impedances by putting them in series across a constant amplitude AC signal. Then measure the voltages across the two impedances.

I made the point that the average X100 amplifier will not generate hundreds of volts just by feeding it with a few volts of drive. More likely, it will generate about 10 volts of distorted output.

That seems fair comment.

 

[meBigGuy]

I just re-read the OP. I assume you have a function generator and a separate 100X amplifier that already gives you the high voltage you need. Are you saying you need different output impedances to get the measurements you need, not just different drive voltages.

1. If yes, what range of impedences do you need and how many steps?

2. If no, why not a fixed output resistance and a pot on the input to the amplifier.

I'm still not clear what your specs are with respect to drive voltage and drive impedance.

 

[vk6kro]

The OP was fairly garbled but it asked for a design of a suitable amplifier.

There was no description of the actual type of piezo being tested. It could be some kind of filter or an echo sounder transducer for a boat, or a tweeter for a sound system.

So, rather than tackle the difficult high voltage amplifier, it is easier to use a 10 cent resistor to measure the impedance.

 

[EE4life]

The input impedance ranges from 50 to 50kohms. Drive voltage from 0.5 voltage to 100 V.

In piezoelectric materials, the impedance depends strongly on the driving voltage and the output vibration levels and the frequency. My aim is to test the impedance response using constant vibration conditions, which would require larger and larger voltages as you move farther away from resonance.

Please understand my question: I am asking how I can administer high voltages and low voltages in the same output. I sometimes need high voltages, so I cannot use an op amp, rather I am using a high speed power amplifier to amplify the signal. I need good sensitivity at the lower end, but I do not want to use two separate systems, so I use a pot between the func gen and the amp.

I believe the suggestion of using a relay between two function generators, one for high voltage and one for lower voltage, is the ideal solution. When I want to switch from high voltage to low voltage, I can bring down the voltage from the high power, then switch seamlessly to the low voltage func gen by matching the driving voltage.

Or I can use a MOSFET between the function generator and the amp. I can change gate voltage to change the input to the amplifier. What do you think about this? Which MOSFET could I use? Or I could use a digital pot https://www.sparkfun.com/products/10613.

 

[vk6kro]

You seem to be describing a high voltage AC constant current sweep generator.

Assuming you don't have a spare $15000, I can't imagine anyone developing this for you for free.

So, maybe you could describe the results you have been getting so far with standard laboratory equipment. Are you doing this for your own interest or for an employer?

Could you do a manual sweep using a normal signal generator and then plot the results on graph paper?
You can maintain constant current by monitoring the voltage across a small series resistor and varying the voltage to keep this constant.

 

[EE4life]

I use labview to automatically control the voltage to maintain constant current while changing the frequency in steps.

I am doing this as part of my Phd.

We have already built this system and got it to work and also published papers using it, but I am trying to improve it.

My need is this, I need to replace the pot in the diagram of the earlier post with something digital. The goal is to have fine control of the voltage at low voltage, but also have the capacity for high voltage output in the same system. I believe the solutions lie in what I explained in post #21. I am going to try a MOSFET as a variable resistor tomorrow morning and see if I get reasonable results. I will let you know what happens.

 

[EE4life]

It is clear that typical MOSFET's/low level transistors won't work because they are asymmetrical devices, and thus mangle AC sinusoidal signals.

I think we can call this solved. Relays or digital pots should both work. Thank you for all your help and ideas regarding my system and the voltage amplification.

Please reply if you have better ideas. If not, thank you so much for the helpful discussion.